Not sure if I’m posting here for answers or just to document my troubleshooting process and ultimately what I find. But here we go.
I have had a nice and stable farm running for months totaling 6030 plots. Those plots are spread over 3 NAS devices and one JBOD. Each NAS has its own harvester and the JBOD is directly connected to the Full Node. Everything ran fine. Then I bought a 14TB USB drive to do some migration for another project and when I was done with it I figured I would add it to the farm. So I connected it to the full node and added the folder to plots. Soon after, I started getting spurts of notifications from chiadog saying “Seeking plots took too long”. I began to look for the issue and in my troubleshooting I found that one of the drives in my JBOD was spitting out errors. Not a lot, but enough for me to notice in the event viewer. So I ordered a replacement. While waiting for the replacement I got another 18TB USB drive to pull the plots from the drive that would be replaced. Then pointed the node to that drive so those plots could continue to be farmed. Now the system is not even looking at the drive with issues. The “Seeking plots” issue continued. I have since replaced the drive and put the plots back and the ploblem is still there. I get “Seeking plots took too long” messages in spurts. Like over a 8 hour period I will get 100 of them and then it runs smooth for hours.
I am suspecting that adding the USB drives to the same harvester as the JBOD is the issue. The harvester needing to jump from one device to the other which happen to be on different interface types is causing a delay. So to test this, I have just removed the USB drives from the plot folder list. I’m going to let it run this way for a couple days. If it runs clean, I know it is the USB drives (the combination of drive/interface types)
FYI, there is no plotting traffic involved in this setup. Just farming/harvesting.
If anyone has any experience in a similar situation, I’m all ears.